CN102318345B - Moving image encoding apparatus and moving image encoding method - Google Patents

Abstract

A moving image encoding apparatus, which divides moving image data into a plurality of sub-data, encodes the sub-data in parallel by using a plurality of encoders, and thereafter splices the resulting encoded sub-bitstream data into a single bitstream, includes a convergence target deriving function for determining a target value for a first amount of buffer occupancy so that the first amount of buffer occupancy at a point in time at which data corresponding to a last picture contained in first sub-bitstream data is removed from a first hypothetical buffer does not drop below a second amount of buffer occupancy which represents the amount of space that second sub-bitstream data occupies in a second hypothetical buffer at that point in time.

Description

Moving picture encoding device, motion image encoding method

Technical field

Execution mode disclosed herein relates to following moving picture encoding device, motion image encoding method and moving picture encoding computer program: by concurrently the multiple data after dynamic image data is cut apart being encoded, and the generated data amount bitstream data fewer than this dynamic image data at high speed.

Background technology

Dynamic image data generally has very large data volume.Therefore, to send dynamic image data or the in the situation that of will storing dynamic image data in storage device to other devices at the device of processing dynamic image data, by dynamic image data is encoded, its data volume be compressed.Especially, following moving picture encoding device is proposed: in order at high speed dynamic image data to be encoded, dynamic image data is divided into multiple subdatas, use multiple encoders concurrently each subdata to be encoded, then, the each subdata after coding is combined into a bitstream data.

In addition, in the time transmitting the dynamic image data after coding to decoding device, this decoding device must be decoded to the dynamic image data after coding.Therefore, hypothesis decoding device has the buffer memory with specified volume for the dynamic image data after interim memory encoding.In this situation, require dynamic image data code device to converge on all the time the mode in the scope of this specified volume with the data volume of accumulating in this buffer memory, dynamic image data is encoded.Below, the virtual cache this decoding device being had is called standard decoder buffer memory.For example, in Moving PictureExperts Group (MPEG), the idea relevant passing of the occupancy volume occupying with the dynamic image data after coding is defined as to Video Buffeting Verifier (VBV) in standard decoder buffer memory.In VBV, adopt following model: after the dynamic image data after coding is transmitted to bit rate, is accumulated in standard decoder buffer memory with initial delay amount with maximum, at a certain time interval, the data of 1 picture amount of moment taking-up from standard decoder buffer memory.And, in the situation that dynamic image data being encoded with fixed bit rate, the occupancy volume that the coding dynamic image data transmitting to standard decoder buffer memory occupies in standard decoder buffer memory must converge between the maximum permissible value and minimum permissible value of standard decoder buffer memory.

But, in the case of the multiple dynamic image datas in conjunction with encoding separately, even if each coding dynamic image data meets the regulation of standard decoder buffer memory, combine bitstream data after these coding dynamic image datas and sometimes also do not meet the regulation of standard decoder buffer memory.This problem is due to following former thereby generation.That is, be defined as, before the transmission of bitstream data finishes during, this bitstream data all the time with maximum transmit bit rate accumulate in standard decoder buffer memory.But, owing to 2 of continuous transmission coding dynamic image datas being carried out to the timing of combination, the coding dynamic image data combined to rear side carry out in conjunction with before buffer memory amount of recovery, and the amount of recovery of buffer memory in the time that binding site is processed as not divided continuous dynamic image data between, sometimes cannot dynamically obtain matching.Therefore,, in the time that the last picture of the coding dynamic image data to these 2 front sides in dynamic image data of encoding is decoded, the occupancy volume that the coding dynamic image data of rear side occupies in standard decoder buffer memory and the occupancy volume of bitstream data are inconsistent.

For this problem, following technology is proposed: the result after simulating according to the migration of the occupancy volume to standard decoder buffer memory, initial delay amount in each cut-point that calculating is cut apart, thus, encodes to dynamic image data in the mode of the regulation that meets standard decoder buffer memory.And, following technology is proposed: after the decoding data comprising between the land of the mpeg image data to after cutting apart, so that the occupancy volume of standard decoder buffer memory converges on the mode between maximum permissible value and the minimum permissible value of this buffer memory, the data that comprise between land are encoded again.

Patent documentation 1: TOHKEMY 2008-85673 communique

Patent documentation 2: TOHKEMY 2004-297829 communique

Summary of the invention

In the technology that the migration of buffer memory occupancy volume is simulated, the regulation that represents to violate standard decoder buffer memory in this analog result, need to adjust for determining the parameter for the sendout of the coded-bit of each picture, and repeatedly simulate.Therefore, use the moving picture encoding device of this technology in the coding of dynamic image data is processed, to need the long period.

And, after the interim decoding data comprising between to land, in technology that these decoded data are encoded again, additionally need to be used for carrying out this and again encode the time of processing.And then this technology is not adjusted the buffer memory that the each dynamic image data after cutting apart is carried out to the time point of combination.Therefore, in the inappropriate situation of this setting in conjunction with time point, the coded-bit amount that particular picture is distributed is few, and its result, may produce violent image quality deteriorated.

Therefore, the object of this specification is, the regulation and the moving picture encoding device that can encode to dynamic image data at short notice, motion image encoding method and the dynamic image encoding that meet standard decoder buffer memory are provided.

According to an execution mode, provide a kind of moving picture encoding device.This moving picture encoding device has: storage part, and its storage at least comprises the dynamic image data of the 1st subdata and the 2nd subdata, and the 1st subdata and the 2nd subdata comprise respectively multiple pictures; Handling part, it is by dynamic image data is encoded, the bitstream data that generated data amount is fewer than the data volume of this dynamic image data; And efferent, its output bit flow data.This handling part is achieved as follows function: convergence target export function, it determines the desired value of the 1st occupancy volume, make transmit bitstream data and be accumulated in the buffer memory of this decoder to virtual decoder at the transfer rate with regulation, and when the data suitable with each picture of taking out bitstream data with official hour interval successively from the buffer memory of this decoder and being comprised, taking out from the buffer memory of this decoder encode the 1st time point of the data suitable with last picture that the 1st sub-bitstream data that obtains comprises of the 1st subdata, the 1st occupancy volume that bitstream data occupies in the buffer memory of this decoder be at the 1st time point, the 2nd subdata is encoded occupy in the buffer memory of the 2nd sub-bitstream data at this decoder obtaining the 2nd more than occupancy volume, the 1st encoding function, it determines the sendout of the coded-bit of each picture comprising for the 1st subdata so that the 1st occupancy volume becomes the mode of desired value, according to this sendout, the 1st subdata is encoded, thereby generates the 1st sub-bitstream data, the 2nd encoding function, it generates the 2nd sub-bitstream data by the 2nd subdata is encoded, and binding function, it generates bitstream data by the 1st sub-bitstream data and the 2nd sub-bitstream data are carried out to combination.

And, according to other execution modes, a kind of motion image encoding method is provided, this motion image encoding method includes respectively the 1st subdata of multiple pictures and the dynamic image data of the 2nd subdata is encoded at least comprising, thus the generated data amount bitstream data fewer than the data volume of this dynamic image data.In this motion image encoding method, determine the desired value of the 1st occupancy volume, make transmit bitstream data and be accumulated in the buffer memory of this decoder to virtual decoder at the transfer rate with regulation, and when the data suitable with each picture of taking out bitstream data with official hour interval successively from the buffer memory of this decoder and being comprised, taking out from the buffer memory of this decoder encode the 1st time point of the data suitable with last picture that the 1st sub-bitstream data that obtains comprises of the 1st subdata, the 1st occupancy volume that bitstream data occupies in the buffer memory of this decoder be at the 1st time point, the 2nd subdata is encoded occupy in the buffer memory of the 2nd sub-bitstream data at this decoder obtaining the 2nd more than occupancy volume, so that the 1st occupancy volume becomes the mode of desired value, determine the sendout of the coded-bit of each picture comprising for the 1st subdata, according to this sendout, the 1st subdata is encoded, thereby generate the 1st sub-bitstream data, by the 2nd subdata is encoded, generate the 2nd sub-bitstream data, by the 1st sub-bitstream data and the 2nd sub-bitstream data are carried out to combination, generate bitstream data.

And then, according to other execution modes, a kind of moving picture encoding computer program is provided, this moving picture encoding includes respectively the 1st subdata of multiple pictures and the dynamic image data of the 2nd subdata is encoded with computer program at least comprising, thereby makes the bitstream data that computer generated data amount is fewer than the data volume of this dynamic image data.This computer program comprises the order that makes computer carry out following steps: the desired value that determines the 1st occupancy volume, make transmit bitstream data and be accumulated in the buffer memory of this decoder to virtual decoder at the transfer rate with regulation, and when the data suitable with each picture of taking out bitstream data with official hour interval successively from the buffer memory of this decoder and being comprised, taking out from the buffer memory of this decoder encode the 1st time point of the data suitable with last picture that the 1st sub-bitstream data that obtains comprises of the 1st subdata, the 1st occupancy volume that bitstream data occupies in the buffer memory of this decoder be at the 1st time point, the 2nd subdata is encoded occupy in the buffer memory of the 2nd sub-bitstream data at this decoder obtaining the 2nd more than occupancy volume, so that the 1st occupancy volume becomes the mode of desired value, determine the sendout of the coded-bit of each picture comprising for the 1st subdata, according to this sendout, the 1st subdata is encoded, thereby generate the 1st sub-bitstream data, by the 2nd subdata is encoded, generate the 2nd sub-bitstream data, by the 1st sub-bitstream data and the 2nd sub-bitstream data are carried out to combination, generate bitstream data.

By the key element and the combination that particularly point out in claims, realize and reach objects and advantages of the present invention.

Hope understands, above-mentioned general description and following detailed description are exemplary and illustrative, and unlike claims, limit the present invention.

The disclosed moving picture encoding device of this specification, motion image encoding method and dynamic image encoding meet the regulation of standard decoder buffer memory, and can encode to dynamic image data at short notice.

Brief description of the drawings

Fig. 1 is the summary construction diagram of the moving picture encoding device of an execution mode.

Fig. 2 is the functional block diagram that the handling part of the moving picture encoding device of the function realizing in order to carry out the processing that dynamic image data is encoded is shown.

Fig. 3 (a) is the figure of an example of the migration of the occupancy volume that illustrates that 1 sub-bitstream data occupies in standard decoder buffer memory.Fig. 3 (b) is the figure of an example of the migration of the occupancy volume that illustrates that 2 combined sub-bitstream data occupy respectively in standard decoder buffer memory.Fig. 3 (c) illustrates the figure that 2 sub-bitstream data is combined into an example of the migration of the occupancy volume that the bitstream data after 1 occupies in standard decoder buffer memory.

Fig. 4 is the figure of an example of the migration of the occupancy volume that illustrates that the sub-bitstream data of the rear side near 2 combined sub-bitstream data of the migration moment occupies in standard decoder buffer memory.

Fig. 5 is the functional block diagram of coding portion.

Fig. 6 (a) is the curve chart that an example of the migration of the lower limit of the occupancy volume of standard decoder buffer memory is shown.And Fig. 6 (b) is the curve chart that an example of the migration of the higher limit of the occupancy volume of standard decoder buffer memory is shown.

Fig. 7 is illustrated in the migration of occupancy volume that occupies in standard decoder buffer memory of bitstream data and the figure of the corresponding relation of 2 combined sub-bitstream data that this bitstream data comprises in the migration moment.

Fig. 8 is the figure illustrating by the motion flow of the coding processing of the dynamic image data of the computer program control of carrying out on the handling part of moving picture encoding device.

Fig. 9 is the figure that the motion flow of the coding processing of the subdata of being undertaken by the each coding portion of the computer program control of carrying out on the handling part of moving picture encoding device is shown.

Label declaration

1: moving picture encoding device; 11: storage part; 12: handling part; 13: input and output portion; 21: control part; 22: cutting part; 23-1,23-2 ..., 23-n: coding portion; 24-1,24-2 ..., 24-(n-1): convergence target leading-out portion; 25-1,25-2 ..., 25-(n-1): joint portion; 31: Data Management Department; 32: orthogonal conversion/quantization unit; 33: variable length code portion; 34: the buffer memory upper limit/lower limit control part; 35: amount of information control part.

Embodiment

Below, with reference to the accompanying drawings of the moving picture encoding device of an execution mode.

Dynamic image data is divided into multiple subdatas by this moving picture encoding device, and the sub-bitstream data that uses different coding device to encode according to each subdata and generate is combined into a bitstream data.This moving picture encoding device control is for the coded-bit sendout of the sub-bitstream data of the front side in 2 combined sub-bitstream data, to meet the regulation of standard decoder buffer memory.And then this moving picture encoding device, between the sub-bitstream data of front side and the sub-bitstream data of rear side, inserts the invalid bit column of taking out in the lump when its last picture is decoded.Thus, this moving picture encoding device has been eliminated at last picture that the sub-bitstream data of front side is comprised and has been carried out the poor of occupancy volume that the sub-bitstream data of occupancy volume that the time point bitstream data of decoding occupies in standard decoder buffer memory and rear side occupies in standard decoder buffer memory.

In addition, the picture comprising as the dynamic image data of coded object can be the either party of the field obtaining with interlace mode or the frame of obtaining with progressive mode.

Fig. 1 is the summary construction diagram of the moving picture encoding device 1 of an execution mode.As shown in Figure 1, moving picture encoding device 1 has storage part 11, handling part 12, input and output portion 13.

Storage part 11 for example has at least any one party in semiconductor memory, disk set or optical disc apparatus.And storage part 11 is stored in the computer program and the various data that use in moving picture encoding device 1.And storage part 11 is stored the dynamic image data as coded object.And storage part 11 also can be stored the bitstream data of by handling part 12, dynamic image data being encoded and generate.

Input and output portion 13 for example has communication interface and the control circuit thereof for moving picture encoding device 1 and communication network (not shown) are coupled together.And input and output portion 13 exports the bitstream data of by handling part 12, dynamic image data being encoded and generate via communication network to other devices.And input and output portion 13 also can obtain the dynamic image data as coded object from other devices via communication network, and this dynamic image data is handed to handling part 12.

Handling part 12 has one or more processors, memory and peripheral circuit thereof.And the dynamic image data that handling part 12 is obtained to the dynamic image data reading in from storage part 11 or via input and output portion 13 is encoded.

Fig. 2 is the functional block diagram that the handling part 12 of the function realizing for dynamic image data is encoded is shown.As shown in Figure 2, handling part 12 have control part 21, cutting part 22, n the 23-1 of coding portion, 23-2 ..., 23-n, (n-1) individual convergence target leading-out portion 24-1,24-2 ..., 24-(n-1), (n-1) individual joint portion 25-1,25-2 ..., 25-(n-1).Wherein, n is more than 2 integer.

These each portions that handling part 12 has are functional modules that the computer program carried out on the processor by having at handling part 12 is installed.These each portions that handling part 12 has also can be respectively as independent computing circuit or as the integrated circuit these computing circuits being carried out after integrated, be arranged on moving picture encoding device 1.

It is all that control part 21 is controlled handling part 12.Therefore, control part 21 to each coding 23-m of portion (wherein m=1,2 ..., n) input is for determining for the parameter of coded-bit sendout of subdata that is input to this coding portion each coding 23-m of portion.And, control part 21 to each convergence target leading-out portion 24-k (wherein k=1,2 ..., n-1) input is for making each convergence target leading-out portion 24-k determine the parameter of the desired value of the occupancy volume that bitstream data occupies at standard decoder buffer memory.In addition, about these parameters, with the 23-m of coding portion narration in the back together with convergence target leading-out portion 24-k.

And control part 21 reads in as the dynamic image data of coded object or via input and output portion 13 and obtains the dynamic image data as coded object from storage part 11, and this dynamic image data is handed to handling part 12.Now, in the case of obtaining together the information relevant with the content of this dynamic image data with dynamic image data, also this information can be handed to handling part 12.And control part 21 also can read in multiple dynamic image datas or obtain multiple dynamic image datas via input and output portion 13 from storage part 11, and these multiple dynamic image datas are handed to cutting part 22.And then handling part 12 also can be encoded to cutting apart in advance the multiple dynamic image datas that obtain by the independent 23-m of coding portion, in conjunction with the sub-bitstream data that each dynamic image data is encoded and obtained, thus, generate a bitstream data.In this situation, also can omit cutting part 22.And these multiple dynamic image datas are directly inputted to each coding 23-m of portion by control part 21.

The number n of the 23-m of coding portion that cutting part 22 has according to handling part 12, cuts apart the dynamic image data of accepting from control part 21, thereby generates subdata.For example, cutting part 22 is divided into dynamic image data n in time equably, thereby generates n subdata.And in the case of accepting the information relevant with the content of dynamic image data from control part 21, cutting part 22 also can make according to this information the time span difference of each subdata.For example, cutting part 22 also can so that process the mode of needed time equalization for the coding of each subdata, be cut apart dynamic image data.Therefore, for example, cutting part 22 also can, with reference to the information relevant with the content of dynamic image data, be processed the mode of the time span of the subdata of the relatively short picture getting final product of needed time to extend the coding that comprises a large amount of static scenes etc., dynamic image data is cut apart.On the contrary, cutting part 22 also can be processed to shorten the coding that comprises the more scene of a large amount of motions etc. the mode of the time span of the subdata of the picture that the needed time relatively grows, and dynamic image data is cut apart.And then the data that the certain length in dynamic image datas is taken out at multiple positions that cutting part 22 also can be from dynamic image data, using each data of taking out as the subdata that will encode.And then in the situation that accepting multiple dynamic image data from control part 21, cutting part 22 also can, with the number n of the 23-m of coding portion, be cut apart the dynamic image data combining after these multiple dynamic image datas, generate subdata.

And the mode that cutting part 22 also can be taking the number of the picture that each subdata was comprised as the multiple of Group Of Pictures (GOP), adjusts the length of subdata.In addition, GOP is the structure of regulation with the picture group that comprises continuous multiple pictures of some cycles repetition, and in GOP, regulation is for coding method between each picture application of frame intra coding method or application of frame.In addition, inner frame coding method is the method that only uses the information comprising in a pictures of coded object to encode to this picture.On the other hand, interframe encode method is the method that uses the picture of coded object and the information of front and back picture thereof to encode to the picture of coded object.

Cutting part 22 according to reproduce time time sequencing, to the 23-1 of coding portion, 23-2 ..., either party in 23-n inputs each subdata generating.And cutting part 22 is inputted the quantity of the picture that each subdata comprises to the coding portion of this subdata of input.

Convergence target leading-out portion 24-1,24-2 ..., 24-(n-1) calculates respectively the relevant desired value of the occupancy volume with standard decoder buffer memory of the time point that last picture that the sub-bitstream data of the front side in 2 combined sub-bitstream data is comprised decodes.In addition, convergence target leading-out portion 24-1,24-2 ..., 24-(n-1) has identical 26S Proteasome Structure and Function, so, below a convergence target leading-out portion 24-m (1≤m≤n-1) is described.

Thereby generate 1 bitstream data if 2 sub-bitstream data are carried out to combination, and transmit this bitstream data to standard decoder buffer memory.In this situation, the occupancy volume preferably occupying in standard decoder buffer memory in this bitstream data of decoding time point of the last picture of the sub-bitstream data of front side equates with the occupancy volume that the sub-bitstream data of rear side before this time point occupies in standard decoder buffer memory.If meet this condition, only generate each sub-bitstream data in the mode of the regulation of adherence to standard decoder buffer, occupancy volume can be higher than the maximum permissible value of standard decoder buffer memory, and can be lower than the minimum permissible value of standard decoder buffer memory.In addition, below, the decoding moment of the last picture of the sub-bitstream data of the front side in 2 combined sub-bitstream data is called to the migration moment.And the time point that the sub-bitstream data of the rear side in 2 combined sub-bitstream data is started to be accumulated in standard decoder buffer memory was called in conjunction with the moment.

Use Fig. 3 (a)～Fig. 3 (c) that this situation is described.Fig. 3 (a) illustrates 1 sub-bitstream data SBS _mthe figure of one example of the migration of the occupancy volume occupying in standard decoder buffer memory.And Fig. 3 (b) illustrates 2 combined sub-bitstream data SBS _m, SBS _m+1the figure of one example of the migration of the occupancy volume occupying in standard decoder buffer memory respectively.And then Fig. 3 (c) is 2 sub-bitstream data SBS that illustrate shown in Fig. 3 (b) _m, SBS _m+1be combined into the figure of an example of the migration of the occupancy volume that the bitstream data BS after 1 occupies in standard decoder buffer memory.

In Fig. 3 (a)～Fig. 3 (c), transverse axis represents the time, and the longitudinal axis represents the occupancy volume of standard decoder buffer memory.And line 301 represents the maximum permissible value of the occupancy volume of standard decoder buffer memory.And, moment t ₀represent sub-bitstream data SBS _mstart to be accumulated in the moment in standard decoder buffer memory.And then, t during this time _drepresent initial buffer delay amount, t during this time _f0represent the picture interval determined by frame per second, for example during t _f0it is 1/30 second.And in Fig. 3 (a) and Fig. 3 (b), curve 310 represents sub-bitstream data SBS _mthe migration of the occupancy volume occupying in standard decoder buffer memory.And in Fig. 3 (b) and Fig. 3 (c), curve 320 represents and sub-bitstream data SBS _min conjunction with next sub-bitstream data SBS _m+1the migration of the occupancy volume occupying in standard decoder buffer memory.And then in Fig. 3 (c), curve 330 represents the migration of the occupancy volume that bitstream data BS occupies in standard decoder buffer memory.And, in Fig. 3 (b) and Fig. 3 (c), moment t _trrepresent the sub-bitstream data SBS to front side _mlast picture migration moment of decoding, moment t _jrepresent the sub-bitstream data SBS of rear side _m+1start to be accumulated in the combination moment in standard decoder buffer memory.

As shown in Fig. 3 (a), from moment t ₀play through initial buffer delay amount t _dtill, sub-bitstream data SBS _mthe occupancy volume time per unit occupying in standard decoder buffer memory transmits bit rate with maximum to be increased.Then, from moment t ₀rise and passed through initial buffer delay amount t _dtime point, from standard decoder buffer memory, take out and be equivalent to the data d of initial picture ₁.Therefore, occupancy volume reduces d ₁.After this time point, every through picture interval t _f0time, from standard decoder buffer memory, take out the data of a pictures amount.And, transmitting sub-bitstream data SBS to standard decoder buffer memory _mprocess in, taking out after the data of a pictures amount to during taking out the data of next picture amount, occupancy volume time per unit transmits bit rate increase with maximum.On the other hand, finishing to transmit sub-bitstream data SBS to standard decoder buffer memory _mmoment t _eafter, occupancy volume does not increase, every through picture interval t _f0time, the data of minimizing one pictures amount.Then,, in the time that taking-up is equivalent to the data of last picture, occupancy volume becomes 0.

And, as shown in Fig. 3 (b), after in conjunction with moment tj, sub-bitstream data SBS _m+1follow sub-bitstream data SBS _mand be accumulated in standard decoder buffer memory.In this situation, according to the regulation of standard decoder buffer memory, be considered as at moment t _ealso transmit bitstream data BS to standard decoder buffer memory continuously later.Therefore, as shown in dotted line 311, at moment t _eafter, after a pictures is decoded till during next pictures is decoded, the occupancy volume that bitstream data BS occupies in standard decoder buffer memory is considered as time per unit and transmits bit rate increase with maximum.On the other hand, as shown in curve 320, in conjunction with moment t _jafter, at process initial delay amount t _dbefore, sub-bitstream data SBS _m+1the occupancy volume time per unit occupying in standard decoder buffer memory transmits bit rate with maximum to be increased.But, if moment t _elater decoding be equivalent to sub-bitstream data SBS _mthe data of picture less, move moment t _trthe residual occupancy volume S of bitstream data BS _rthan sub-bitstream data SBS _m+1in standard decoder buffer memory, occupy migration time occupancy volume S _trmany.

Therefore, as shown in Fig. 3 (c), at migration moment t _trafter, the occupancy volume that bitstream data BS occupies in standard decoder buffer memory becomes, antithetical phrase bitstream data SBS _m+1the occupancy volume occupying in standard decoder buffer memory adds residual occupancy volume S _roccupancy volume S during with migration _trthe absolute value delta S of difference _rand the value obtaining.On the other hand, sometimes with at moment t arbitrarily _nsub-bitstream data SBS _m+1the occupancy volume occupying in standard decoder buffer memory is near the mode maximum permissible value of standard decoder buffer memory, generates sub-bitstream data SBS _m+1.In this situation, antithetical phrase bitstream data SBS _m+1the occupancy volume occupying in standard decoder buffer memory adds Δ S _r, thus, the be above standard maximum permissible value of decoder buffer of the occupancy volume that bitstream data BS occupies in standard decoder buffer memory.

On the contrary, if moment t _elater decoding be equivalent to sub-bitstream data SBS _mthe data of picture more, move moment t _trthe residual occupancy volume S of bitstream data BS _rthan sub-bitstream data SBS _m+1migration time occupancy volume S _trfew.In this situation, at migration moment t _trafter, the occupancy volume that bitstream data BS occupies in standard decoder buffer memory becomes, from sub-bitstream data SBS _m+1in the occupancy volume occupying, deduct residual occupancy volume S in standard decoder buffer memory _roccupancy volume S during with migration _trthe absolute value delta S of difference _rand the value obtaining.On the other hand, sometimes with at moment neutron bitstream data SBS arbitrarily _m+1the occupancy volume occupying in standard decoder buffer memory is near the mode minimum permissible value of standard decoder buffer memory, generates sub-bitstream data SBS _m+1.In this situation, from sub-bitstream data SBS _m+1in the occupancy volume occupying, deduct Δ S in standard decoder buffer memory _r, thus, the occupancy volume that bitstream data BS occupies in standard decoder buffer memory is lower than the minimum permissible value of standard decoder buffer memory.

Therefore, convergence target leading-out portion 24-m obtain with the sub-bitstream data SBS being generated by the 23-m of coding portion _mthe sub-bitstream data SBS that inscribes when relevant migration, generated by the 23-of coding portion (m+1) _m+1in standard decoder buffer memory, occupy migration time occupancy volume, as a desired value.And convergence target leading-out portion 24-m determines at migration moment antithetical phrase bitstream data SBS _mwith sub-bitstream data SBS _m+1carry out the convergence desired value of the residual occupancy volume occupying in conjunction with the bitstream data BS obtaining in standard decoder buffer memory.If this convergence desired value is below the maximum permissible value of the above and standard decoder buffer memory of when migration occupancy volume.

Occupancy volume and convergence desired value when determining to move, input initial buffer delay amount, maximum transmission bit rate, binding site interframe time from control part 21 to convergence target leading-out portion 24-m.In addition, the binding site interframe time is to play from standard decoder buffer memory and take out and sub-bitstream data SBS from the migration moment _m+1moment of data corresponding to initial picture till during.

Fig. 4 is near the sub-bitstream data SBS that illustrates that the migration moment is _m+1the figure of the migration of the occupancy volume occupying in standard decoder buffer memory.In Fig. 4, moment t _jrepresent in conjunction with the moment moment t _trrepresent the migration moment.And then, moment t ₁represent to take out and sub-bitstream data SBS from standard decoder buffer memory _m+1moment of data corresponding to initial picture.And, t during this time _drepresent initial buffer delay amount, t during this time _frepresent the binding site interframe time.And then curve 401 represents sub-bitstream data SBS _m+1the occupancy volume occupying in standard decoder buffer memory.And curve 402 represents the antithetical phrase bitstream data SBS in standard decoder buffer memory _mwith sub-bitstream data SBS _m+1carry out the occupancy volume occupying in conjunction with the bitstream data BS obtaining in standard decoder buffer memory.

Here, according to the definition of standard decoder buffer memory, from moment t _jto moment t ₁till during be accumulated in the sub-bitstream data SBS in standard decoder buffer memory _m+1occupancy volume become, at initial buffer delay amount t _din be multiplied by the maximum value that transmits bit rate and obtain.And, from migration moment t _trto moment t ₁till during be accumulated in the sub-bitstream data SBS in standard decoder buffer memory _m+1occupancy volume become, to binding site interframe time t _fbe multiplied by the maximum value that transmits bit rate and obtain.Therefore, as shown in Figure 4, occupancy volume S when convergence target leading-out portion 24-m can be according to following formula computation migration _tr.

[numerical expression 1]

S _tr＝br×(t _d-t _rt) (1)

Here, br is the maximum bit rate that transmits.

Occupancy volume S when convergence target leading-out portion 24-m calculates migration _trafter, occupancy volume S during to migration _tradd the deviation of regulation, thus, calculate convergence desired value.In addition, the deviation of regulation have 0 or on the occasion of.And the deviation of regulation is set to, at migration moment t _tr, take out and sub-bitstream data SBS _mdata corresponding to last picture after, the occupancy volume that bitstream data BS occupies in standard decoder buffer memory is occupancy volume S during for migration _trabove.And the deviation of regulation is set to, at migration moment t _tr, taking out and sub-bitstream data SBS _mdata corresponding to last picture before, the occupancy volume that bitstream data BS occupies in standard decoder buffer memory is below the maximum permissible value of standard decoder buffer memory.Therefore, the deviation of regulation for example can become the number of coded bits that obtains by a pictures or GOP the are encoded value of 3 times with respect to the standard deviation of the skew of allocation bit number.

While restraining target leading-out portion 24-m by migration, occupancy volume is handed to joint portion 25-m.And convergence desired value is handed to the 23-m of coding portion by convergence target leading-out portion 24-m.

The 23-1 of coding portion, 23-2 ..., 23-n encodes to inputted subdata, thereby generates sub-bitstream data SBS ₁, SBS ₂..., SBS _n.In addition, the 23-1 of coding portion, 23-2 ..., 23-n has identical 26S Proteasome Structure and Function, so, below a 23-m of coding portion is described.

Fig. 5 is the functional block diagram of the 23-m of coding portion.As shown in Figure 5, the 23-m of coding portion has Data Management Department 31, orthogonal conversion/quantization unit 32, variable length code portion 33, the buffer memory upper limit/lower limit control part 34, amount of information control part 35.

Data Management Department 31, according to the order of the picture being encoded, hands to orthogonal conversion/quantization unit 32 by the subdata of accepting from cutting part 22 in the mode of a sheet by a sheet picture.And Data Management Department 31 also represents to 32 inputs of orthogonal conversion/quantization unit the signal that carries out interframe encode and still carry out intraframe coding handing to the picture of orthogonal conversion/quantization unit 32.

And then it is photo current numbering that Data Management Department 31 inputs as the coding of the picture of present encoding handling object to the buffer memory upper limit/lower limit control part 34, to determine higher limit and the lower limit for the occupancy volume of the standard decoder buffer memory of each picture.

Orthogonal conversion/quantization unit 32 carries out orthogonal conversion to the picture of accepting from Data Management Department 31, and the frequency signal obtaining by this orthogonal conversion process is quantized.Thus, orthogonal conversion/quantization unit 32 generates the few quantized signal of bit number that bit number has than original picture.Therefore, orthogonal conversion/quantization unit 32 for example also can, according to MPEG-2, MPEG-4 or the either party in the various moving picture encoding standards of Video Coding (H.264MPEG-4AVC) etc. H.264MPEG-4Advanced, be carried out orthogonal conversion process and quantification treatment.

As an example of orthogonal conversion process and quantification treatment, first, orthogonal conversion/quantization unit 32 is divided into the photo current of accepting from Data Management Department 31 to have multiple of stated number pixel.Below this piece is called to macro block.And macro block for example comprises 16 × 16 pixels.

The calculus of differences that orthogonal conversion/quantization unit 32 is carried out between each macro block and predicted picture.Then, orthogonal conversion/quantization unit 32 generates the difference value corresponding with each pixel in the macro block obtaining by this calculus of differences, as predictive error signal.Now, orthogonal conversion/quantization unit 32 is according to the signal that represents the photo current of accepting from Data Management Department 31 to carry out interframe encode and still carry out intraframe coding, selects the either party in predicted picture that predicted picture that interframe encode uses or intraframe coding use.In addition, as described later, the predicted picture that interframe encode is used is to generate according to encoded picture.On the other hand, the predicted picture that intraframe coding is used is to generate according to the encoded macro block of photo current.

Orthogonal conversion/quantization unit 32, by the predictive error signal of each macro block is carried out to orthogonal conversion, is obtained and is represented the frequency content of horizontal direction of predictive error signal and the frequency signal of the frequency content of vertical direction.For example, as orthogonal conversion process, orthogonal conversion/quantization unit 32 is carried out discrete cosine transform (Discrete CosineTransform, DCT) to predictive error signal, thus, obtains the group of the DCT coefficient of each macro block, as frequency signal.

Then, orthogonal conversion/quantization unit 32, according to the quantization parameter being determined by amount of information control part 35, quantizes frequency signal.This quantification treatment is to utilize a signal value to represent the processing of certain interval signal value comprising.And this certain interval is called as quantization amplitude.For example, orthogonal conversion/quantization unit 32 is cast out the next bit of the stated number that is equivalent to quantization amplitude from frequency signal, thus this frequency signal is quantized.Quantization amplitude is determined by quantization parameter.For example, orthogonal conversion/quantization unit 32 is the function for the value of the quantization amplitude of the value of quantization parameter according to expression, determines the quantization amplitude using.And this function can be the increasing function for the value of quantization parameter, be predefined.Or, prepare in advance multiplely for specifying and the frequency content of horizontal direction and the vertical direction quantization matrix of corresponding quantization amplitude respectively, be stored in storage part 11.Then, orthogonal conversion/quantization unit 32, according to quantization parameter, selects to be stored in the specific quantization matrix in these quantization matrixes in storage part 11.Then, orthogonal conversion/quantization unit 32, with reference to the quantization matrix of selecting, determines the quantization amplitude for each frequency content of frequency signal.In this situation, orthogonal conversion/quantization unit 32 selects the value of quantization parameter larger, for the larger quantization matrix of the quantization amplitude of each frequency content.

Orthogonal conversion/quantization unit 32 is carried out quantification treatment, thus, can cut down the amount of bits of the each frequency content for representing frequency signal, so, can reduce the amount of information that each macro block comprises.Orthogonal conversion/quantization unit 32 provides quantized signal to variable length code portion 33.

And for generation forecast image, orthogonal conversion/quantization unit 32 is multiplied by the stated number of the quantization amplitude that is equivalent to be determined by quantization parameter in quantized signal, carry out thus re-quantization.By this re-quantization, the frequency signal to each macro block, the group of for example DCT coefficient are restored.Then, orthogonal conversion/quantization unit 32 carries out contrary orthogonal conversion process to frequency signal.For example, the in the situation that of carrying out DCT processing in orthogonal converter section 22, orthogonal conversion/quantization unit 32 is carried out inverse DCT processing for inverse quantized signal.By quantized signal is carried out, re-quantization is processed and contrary orthogonal conversion process, reproduce have with coding before the predictive error signal of information of predictive error signal same degree.

Orthogonal conversion/quantization unit 32, for the picture after interframe encode, adds the predictive error signal after the reproduction corresponding with this pixel to each pixel value of the predicted picture after motion compensation described later.On the other hand, orthogonal conversion/quantization unit 32, for the picture after intraframe coding, adds the predictive error signal after the reproduction corresponding with this pixel to each pixel value of the predicted picture generating according to encoded macro block.Carry out these for each macro block and process, thus, orthogonal conversion/quantization unit 32 obtains the predicted pictures for photo current.

Interim Storage Estimation picture in the memory that orthogonal conversion/quantization unit 32 has at handling part 12, as new reference picture.Then, orthogonal conversion/quantization unit 32 utilizes this reference picture in the time of generation forecast image.In addition, orthogonal conversion/quantization unit 32 is stored the reference picture of predetermined regulation number, in the time that the number of reference picture exceedes this regulation number, abandons in order old reference picture.

The predicted picture of using in order to generate interframe encode, orthogonal conversion/quantization unit 32 uses each macro block and the encoded reference picture of photo current, obtains motion vector.The piece of the macro block that motion vector represents photo current and the reference picture the most similar to this macro block amount of movement spatially.For example, the concern macro block that orthogonal conversion/quantization unit 32 is carried out photo current mates with the piece of each reference picture, thus, determines the region in reference picture and this reference picture the most consistent with macro block from each reference picture.Then, orthogonal conversion/quantization unit 32 is using the vector respectively with following key element as motion vector: the macro block position of photo current and the region the most consistent with this macro block are in the horizontal direction with the amount of movement of vertical direction and represent the identifying information of the reference picture under this region.

Each macro block that orthogonal conversion/quantization unit 32 comprises for photo current, obtains respectively motion vector.Then, orthogonal conversion/quantization unit 32 carries out motion compensation according to the motion vector of obtaining to reference picture, thus, generates the predicted picture of the block unit after motion compensation.In addition, motion compensation is following processing: in order to offset the position offset of piece of the macro block that represented by motion vector and the reference picture the most similar to this macro block, the position of the piece of mobile this most similar reference picture.

And orthogonal conversion/quantization unit 32 is for the concern macro block in photo current, carry out interpolation from the pixel value comprising with the left side of this concern macro block or the adjacent encoded macro block of upside, generate the predicted picture that intraframe coding is used.

Variable length code portion 33 encodes to quantized signal and the motion vector accepted from orthogonal conversion/quantization unit 32, thus, has generated for original picture compression the group of the bit after data volume.Then, the execution sequence that variable length code portion 33 processes according to coding, the group of the bit that connecting needle generates the each picture that is input to the subdata of the 23-m of coding portion and comprises, thus, generates the sub-bitstream data corresponding with this subdata.Therefore, for example, variable length code portion 33 is for this quantized signal, carry out distribute the higher signal value of probability of happening shorter, the variable length code processing of the code word of variable-length.For example, variable length code portion 33 can carry out Huffman encoding processing or arithmetic coding processing, as variable length code processing.

Variable length code portion 33 exports the sub-bitstream data generating to joint portion 25-m.And, the coded-bit length that variable length code portion 33 distributes to the actual each picture that generated sub-bitstream data is comprised of amount of information control part 35 notice, each macro block or each GOP.

The buffer memory upper limit/lower limit control part 34 determines the limits value of the occupancy volume that bitstream data occupies in standard decoder buffer memory.Particularly, the buffer memory upper limit/lower limit control part 34 determines from standard decoder buffer memory to take out the lower limit of occupancy volume time point, standard decoder buffer memory of the coded data corresponding with photo current as coded object.And the buffer memory upper limit/lower limit control part 34 determines to accumulate the higher limit of the occupancy volume of the standard decoder buffer memory of the time point of the coded data corresponding with photo current as coded object in standard decoder buffer memory.

In order to determine higher limit and lower limit, from control part 21 to buffer memory, the upper limit/lower limit control part 34 is inputted initial setting higher limit and the initial setting lower limit of standard decoder buffer memory and is input to the picture number that the subdata of the 23-m of coding portion comprises.And 34 inputs of the upper limit/lower limit control part are as the numbering of the photo current of coding handling object from Data Management Department 31 to buffer memory.And then, input the convergence desired value in the migration moment that last picture that the sub-bitstream data to being generated by the 23-m of coding portion comprises decodes from convergence target leading-out portion 24-m to the buffer memory upper limit/lower limit control part 34.

The buffer memory upper limit/lower limit control part 34 is for combining the sub-bitstream data SBS being generated by the 23-m of coding portion, 23-(m+1) respectively _m, SBS _m+1after the occupancy volume that occupies in standard decoder buffer memory of bitstream data BS, preset lower limit.Particularly, the buffer memory upper limit/lower limit control part 34 as the lower limit of making decision: from sub-bitstream data SBS _mto SBS _m+1in the migration moment of migration, the occupancy volume that bitstream data BS occupies in standard decoder buffer memory is not less than convergence desired value.And then the buffer memory upper limit/lower limit control part 34 is set and take out sub-bitstream data SBS from standard decoder buffer memory _mthe lower limit of occupancy volume of bitstream data BS of time point of each picture, the number of coded bits that each picture is distributed can sharply not changed.

For example,, from sub-bitstream data SBS _mbefore starting to be accumulated in control zero hour that elapsed time of rising in standard decoder buffer memory arrives regulation, the buffer memory upper limit/lower limit control part 34 is made as lower limit the initial setting lower limit of inputting from control part 21.In addition, calculate this elapsed time t according to following formula.

[numerical expression 2]

t＝(n _p-1)×f _r+t _d (2)

Here n, _prepresent the numbering of the photo current of the conduct coding handling object of obtaining from Data Management Department 31.And fr represents the number of the picture that subdata of taking out from standard decoder buffer memory per second comprises.And then, t _drepresent from sub-bitstream data SBS _mstart to be accumulated in standard decoder buffer memory rear till take out the initial buffer delay amount of the data corresponding with initial picture.

On the other hand, be greater than and control the zero hour in the situation that at elapsed time t, the buffer memory upper limit/lower limit control part 34 increases progressively lower limit, makes elapsed time t larger, and lower limit approaches convergence desired value.In addition, control the value that is set to more than 0 zero hour and is less than control finish time of regulation.But when only revising when approaching the lower limit of several pictures of subdata end, the bit number that can distribute these pictures is few, sharply deteriorated for the decoding image quality of these pictures.Therefore, preferably control and be set to the zero hour, by due to follow the rising of lower limit of each picture make number of coded bits reduce the image quality degradation inhibiting causing be observer note less than degree.For example, control and be set to the zero hour, elapsed time corresponding to value below the half of the number of the picture comprising with subdata.And, control and can be set as the finish time, from sub-bitstream data SBS _mto sub-bitstream data SBS _m+1the timing of the regulation before the migration moment of migration and after controlling the finish time.But, preferably control the zero hour as far as possible long with the interval of controlling the finish time.By such setup control finish time, the buffer memory upper limit/lower limit control part 34 can make the lower limit of the occupancy volume of standard decoder buffer memory approach convergence desired value, and does not make sharply to reduce for the coded-bit sendout of each picture.

The buffer memory upper limit/lower limit control part 34 for example can be set the lower limit L for photo current according to following formula _np.

[numerical expression 3]

L _np＝L _ini (t _p≤t _start)

L _np＝α(t _p-t _start)

(t _start＜t _p＜t _end) (3)

$\mathrm{\α}=\frac{S_{t\arg \mathrm{et}}-L_{\mathrm{ini}}}{t_{\mathrm{end}}-t_{\mathrm{start}}}$

L _np＝S _target (t _end≤t _p)

Here t, _prepresent the moment corresponding with photo current.And, t _startrepresent to start to revise the control zero hour of lower limit, t _endrepresent to finish to revise the control finish time of lower limit.And then, S _targetrepresent convergence desired value, L _inirepresent initial setting lower limit.

And, suppose that the initial setting higher limit of the standard decoder buffer memory being determined by control part 21 is lower than the situation of convergence desired value.In this situation, the 23-m of coding portion cannot determine the coded-bit sendout for each picture to exceed the mode of this initial setting higher limit.Therefore, the 23-m of coding portion cannot make to take out sub-bitstream data SBS from standard decoder buffer memory _mthe residual occupancy volume of the bitstream data BS of the time point of the data that are equivalent to last picture that comprise approaches convergence desired value.Therefore, in this case, the buffer memory upper limit/lower limit control part 34 is modified to higher limit higher than convergence desired value.For example, the buffer memory upper limit/lower limit control part 34 will be set as the maximum permissible value of the occupancy volume of standard decoder buffer memory for the higher limit of whole pictures.Or the buffer memory upper limit/lower limit control part 34 also can, after control zero hour of regulation and before controlling the finish time, improve the higher limit of taking out the moment of picture from standard decoder buffer memory gradually, until this higher limit is higher than restraining desired value.In addition, the control zero hour relevant with the setting of higher limit and the control relevant with the setting of lower limit the zero hour also can be inconsistent.Equally, the control finish time relevant with the setting of higher limit and the control relevant with the setting of lower limit the finish time also can be inconsistent.But, preferably capping value and lower limit as follows: be provided with all the time the number of coded bits quite poor with at least one pictures amount between higher limit and lower limit.

Fig. 6 (a) is the curve chart that an example of the migration of the lower limit of the standard decoder buffer memory of being set by the buffer memory upper limit/lower limit control part 34 is shown.And Fig. 6 (b) is the curve chart that an example of the migration of the higher limit of the standard decoder buffer memory of being set by the buffer memory upper limit/lower limit control part 34 is shown.

In Fig. 6 (a) and Fig. 6 (b), transverse axis represents the time, and the longitudinal axis represents the occupancy volume of standard decoder buffer memory.And in Fig. 6 (a), line 601 represents the higher limit of the occupancy volume of standard decoder buffer memory, line 602 represents the lower limit of the occupancy volume of standard decoder buffer memory.And then, occupancy volume when dotted line 603 and 604 represents respectively convergence desired value and migration.And curve 605 represents the sub-bitstream data SBS to being generated by the 23-m of coding portion _mwith the sub-bitstream data SBS being generated by the next one coding 23-of portion (m+1) _m+1carry out in conjunction with after the migration of the occupancy volume that occupies in standard decoder buffer memory of bitstream data BS.

As shown in Fig. 6 (a), from sub-bitstream data SBS _mstart to be accumulated in the moment t in standard decoder buffer memory ₀to controlling the t zero hour ₁till during, the lower limit 602 of the occupancy volume of standard decoder buffer memory is set to initial setting lower limit.Then, when process is controlled the t zero hour ₁arrive afterwards and control the t finish time ₂till during, lower limit 602 increases point-blank.Then, controlling the t finish time ₂, lower limit 602 reaches the convergence desired value 603 of more than 604 value of occupancy volume while being set to migration, at moment t ₂keep constant later.Preset lower limit like this, so, at antithetical phrase bitstream data SBS _mthe migration moment t that decodes of last picture _tr, the occupancy volume that bitstream data BS occupies in standard decoder buffer memory is reliably higher than sub-bitstream data SBS _m+1migration time occupancy volume.On the other hand, in this example, the higher limit 601 of the occupancy volume of standard decoder buffer memory is set to, even lower limit 602 is set as restraining desired value 603, occupancy volume 605 can not exceed the abundant high value of this higher limit 601 yet, so the buffer memory upper limit/lower limit control part 34 is not revised higher limit.

In Fig. 6 (b), line 611 represents the higher limit of the occupancy volume of standard decoder buffer memory, and line 612 represents the lower limit of the occupancy volume of standard decoder buffer memory.And then chain-dotted line 613 represents convergence desired value.

As shown in Fig. 6 (b), starting to be accumulated in the moment t standard decoder buffer memory from sub-bitstream data ₀to controlling the t zero hour ₁till during, the lower limit 612 of the occupancy volume of standard decoder buffer memory is set to initial setting lower limit.Then, when process is controlled the t zero hour ₁arrive afterwards and control the t finish time ₂till during, lower limit 612 increases point-blank.Then, controlling the t finish time ₂, lower limit 612 reaches convergence desired value 613.On the other hand, from moment t ₀to controlling the t zero hour ₃till during, the higher limit 611 of the occupancy volume of standard decoder buffer memory is set to initial setting higher limit.Then, when process is controlled the t zero hour ₃arrive afterwards and control the t finish time ₄till during, higher limit 611 increases point-blank.Then, controlling the t finish time ₄, higher limit 611 reaches the maximum permissible value of standard decoder buffer memory, at moment t ₄keep constant later.

The buffer memory upper limit/lower limit control part 34 is notified higher limit and the lower limit for photo current to amount of information control part 35.

Amount of information control part 35 determines quantization parameter, the number of coded bits that this quantization parameter distributes for the picture of controlling being encoded by orthogonal conversion/quantization unit 32.

Therefore, amount of information control part 35 can adopt any means in the determining method of the various quantization parameters less with the value value larger, quantization parameter of target information amount corresponding to number of coded bits that picture is distributed.As this method, amount of information control part 35 can adopt following feedback-type method: the plan according to the number of coded bits that reality is distributed encoded picture as the number of coded bits of target, determines the number of coded bits that next picture is distributed.And amount of information control part 35 also can adopt following feed-forward type method: according to the statistic of the computational chart diagram sheet character such as the motion prediction process in orthogonal conversion/quantization unit 32, determine the number of coded bits that next picture is distributed according to this statistic.And then amount of information control part 35 also can use the method after this feedback-type method and feed-forward type method are combined.

For example, amount of information control part 35, according to the method adopting in the standard testing model 5 of MPEG-2, is obtained the elementary object amount of information of each picture.In this situation, from control part 21 to maximum bit rate and the frame per second of transmitting of amount of information control part 35 inputs.And, represent to be the picture of interframe encode or the picture classification information of the picture of intraframe coding as the photo current of coded object from Data Management Department 31 to 35 inputs of amount of information control part.And then, input higher limit and the lower limit for photo current from the buffer memory upper limit/lower limit control part 34 to amount of information control part 35.Then, amount of information control part 35, according to the parameter of these inputs, is obtained elementary object amount of information, makes in the time transmitting the coded-bit row corresponding with photo current to standard decoder buffer memory, and the occupancy volume of standard decoder buffer memory can not exceed higher limit.And amount of information control part 35, according to the parameter of these inputs, is obtained elementary object amount of information, make take out the coded-bit row corresponding with photo current from standard decoder buffer memory time, the occupancy volume of standard decoder buffer memory can be lower than lower limit.

And then, amount of information control part 35 is obtained the control information amount for each picture, make to restrain desired value in the case of the lower limit of buffer memory occupancy volume is modified to approach from initial setting lower limit, the occupancy volume of standard decoder buffer memory can be lower than this revised lower limit.Then, amount of information control part 35 deducts correction target amount of information from elementary object amount of information, thereby obtains final target information amount.Then, amount of information control part 35 is obtained the quantization parameter corresponding with the value of target information amount.In addition, about the elementary object amount of information relevant with the standard testing model 5 of MPEG-2 and the computational methods of quantization parameter, wish with reference to the URL being determined by http://www.mpeg.org/MPEG/MSSG/tm5/Ch10/Ch10.html.And amount of information control part 35 also can use the rate distortion function of the relation of the disclosed expression target information amount of TOHKEMY 2008-252562 communique and quantization parameter, decides quantization parameter.In this situation, in the case of from the lower limit of initial setting lower limit correction buffer memory occupancy volume, amount of information control part 35 deducts the control information amount for each picture from target information amount, thus, revises this target information amount.

As an example, first, amount of information control part 35 calculates the complexity estimating for photo current.For carry out intraframe coding picture complexity, for undertaken by single directional prediction interframe encode picture complexity and for calculating by following formula respectively by the complexity of the bi-directional predicted picture that carries out interframe encode.Below, the picture that carries out intraframe coding is called to I picture.And, will use the picture that the information of forward picture is carried out interframe encode to be in time called P picture.And then, by use forward picture in time and lean on after both sides' the information of the picture picture that carries out interframe encode be called B picture.

[numerical expression 4]

X _i＝S _iQ _i

X _p＝S _pQ _p (4)

X _b＝S _bQ _b

Here, Xi is the complexity for I picture, and Xp is the complexity for P picture, and Xb is the complexity for B picture.And Si is to be the bit number producing by encoding I picture, for this previous picture at encoded previous picture.Equally, Sp is to be the bit number producing by encoding P picture, for this previous picture at encoded previous picture.And then Sb is to be the bit number producing by encoding B picture, for this previous picture at encoded previous picture.And, Qi, Qp, Qb are respectively the average quantisation parameter using in the situation that previous picture is I picture, P picture, B picture, when previous picture is encoded, and this average quantisation parameter is by the quantization parameter of whole macro blocks is averaged and calculated.In addition, as an example, set as follows the initial value of Xi, Xp, Xb.

Xi＝(160*bitrate)/115

Xp＝(60*bitrate)/115

Xb＝(42*bitrate)/115

Wherein, bitrate is the maximum bit rate that transmits, and is illustrated in the amount of information per second (bit/s) of giving in coding.

Then, amount of information control part 35 is according to complexity Xi, the Xp, the Xb that calculate, calculates for the elementary object amount of information Ti of I picture, for the elementary object amount of information Tp of P picture and for the elementary object amount of information Tb of B picture according to following formula.

[numerical expression 5]

$T_i=\max \{\frac{R}{(1+\frac{N_p{X}_p}{X_i{K}_p}+\frac{N_b{X}_b}{X_i{K}_b})},\frac{\mathrm{bitrate}}{8\×\mathrm{picturerate}}\}$

$T_p=\max \{\frac{R}{{(N}_p+\frac{N_b{K}_p{X}_b}{K_b{X}_p})},\frac{\mathrm{bitrate}}{8\×\mathrm{picturerate}}\}---\left(5\right)$

$T_b=\max \{\frac{R}{{(N}_b+\frac{N_p{K}_b{X}_p}{K_p{X}_b})},\frac{\mathrm{bitrate}}{8\×\mathrm{picturerate}}\}$

Here, Kp, Kb are constants, are generally set as Kp=1.0, Kb=1.4.And R is can be to the surplus of the number of coded bits of GOP distribution.After the coding of each picture completes, upgrade as follows R.

R＝R-Sj

Wherein, if previous picture is I picture, Sj=Si.And, if previous picture is P picture, Sj=Sp.Or, if previous picture is B picture, Sj=Sb.

And, to the initial picture of GOP being I picture while encoding, use the surplus R of the bit number calculating for a upper GOP, upgrade as follows the surplus R of bit number.

R＝G+R

G＝bitrate*N/picturerate

Wherein, N is the number of the picture that comprises of GOP.In addition, for the initial GOP of dynamic image data, R is set as to 0.And then, picturerate be per second in the dynamic image data of coded object during the number (Hz) of picture of interscan.

And in (5) formula, Np, Nb are respectively the uncoded numbers that remains P picture and B picture in the GOP in coded sequence.And function max (a, b) is the function of the value of the larger side in output variable a and b.

And then, amount of information control part 35 uses control information amount Δ T, the target information amount of adjusting the picture of encoding after the control zero hour for the lower limit of the occupancy volume at standard decoder buffer memory, makes the occupancy volume of standard decoder buffer memory can not be less than this lower limit.That is, controlling photo current is encoded after the zero hour in the situation that, amount of information control part 35 calculation correction amount of information Δ T.Then,, if photo current is I picture, amount of information control part 35 will deduct value that control information amount Δ T obtains as the target information amount T for photo current from the elementary object amount of information Ti calculating by (5) formula.And if photo current is P picture, the value that amount of information control part 35 obtains deduct control information amount Δ T from elementary object amount of information Tp is as the target information amount T for photo current.And then if photo current is B picture, the value that amount of information control part 35 obtains deduct control information amount Δ T from elementary object amount of information Tb is as the target information amount T for photo current.

Wherein, preferably control information amount Δ T is the value corresponding to difference of the lower limit with take out last the picture being encoded from standard decoder buffer memory time and the lower limit take out photo current from standard decoder buffer memory time.For example,, in the situation that obtaining lower limit by (3) formula, by following formula calculation correction amount of information Δ T.

[numerical expression 6]

$\mathrm{\ΔT}=\frac{S_{t\arg \mathrm{et}}-L_{\mathrm{ini}}}{N_a}---\left(6\right)$

Here S, _targetrepresent the convergence desired value being calculated by convergence target leading-out portion 24-m, L _inirepresent initial setting lower limit.And, N _abe in the control zero hour of lower limit that starts the occupancy volume of adjusting standard decoder buffer memory from the buffer memory upper limit/lower limit control part 34 till the number of the picture of encoding in during finishing to adjust control finish time of this lower limit.In addition, amount of information control part 35 also can be set as 0 by the control information amount Δ T of I picture larger the coding impact on other pictures.In this situation, in order to determine the control information amount Δ T for P picture and B picture, as long as make the N in (6) formula _afor the total at the number of the P picture of encoding in during the finish time from controlling the zero hour till controlling and B picture.

Then, amount of information control part 35 determines quantization parameter according to target information amount T.Therefore, amount of information control part 35 calculates the sufficient degree d of the standard decoder buffer memory before j macro block encoded _ij, d _pj, d _bj.In addition, sufficient degree d _ijcorresponding to I picture, sufficient degree d _pjcorresponding to P picture, sufficient degree d _bjcorresponding to B picture.Calculate this abundance degree d according to following formula _ij, d _pj, d _bj.

[numerical expression 7]

$d_{\mathrm{ij}}=d_{i0}+B_{j-1}-\left(\frac{T_i\×(j-1)}{\mathrm{MBN}}\right)$

$d_{\mathrm{pj}}=d_{p0}+B_{j-1}-\left(\frac{T_p\×(j-1)}{\mathrm{MBN}}\right)---\left(7\right)$

$d_{\mathrm{bj}}=d_{b0}+B_{j-1}-\left(\frac{T_b\×(j-1)}{\mathrm{MBN}}\right)$

Here B, _j-1be by the 1st to (j-1) individual whole macro blocks bit number generating of encoding.MBN is the quantity of the macro block that comprises of picture.And, d _i0, d _p0, d _b0it is the initial value of sufficient degree.In addition, about particular picture, sufficient degree d when last macro block is encoded _iMBN, d _pMBN, d _bMBNbecome the initial value d for the sufficient degree of the picture of next same type _i0, d _p0, d _b0.

Finally, amount of information control part 35 uses this abundance degree d _ij, d _pj, d _bj, determine the quantization parameter for j macro block by following formula.

[numerical expression 8]

$Q_j=\left(\frac{d_j\×31}{r}\right)$

$r=2\×\frac{\mathrm{bitrate}}{\mathrm{picturerate}}---\left(8\right)$

Wherein, if photo current is I picture, d _jfor d _ijif photo current is P picture, d _jfor d _pjif photo current is B picture, d _jfor d _bj.

The quantization parameter of obtaining is handed to orthogonal conversion/quantization unit 32 by amount of information control part 35.And amount of information control part 35 notifies basis for the sub-bitstream data SBS accepting from variable length code portion 33 to joint portion 25-m _mthe migration moment that the actual coding bit number of the each picture comprising calculates, comprise sub-bitstream data SBS _mthe residual occupancy volume of bitstream data BS.In order to obtain this residual occupancy volume, amount of information control part 35 is to from sub-bitstream data SBS _mafter starting to be accumulated in standard decoder buffer memory, arrive antithetical phrase bitstream data SBS _mlast picture decode till during length, be multiplied by the maximum bit rate that transmits.Then, amount of information control part 35 deducts sub-bitstream data SBS from this multiplication value _mthe total of the number of coded bits of the whole pictures that comprise, thus, calculates residual occupancy volume.

Joint portion 25-1,25-2 ..., 25-(n-1) respectively according to reproduce time sequencing, in conjunction be input to joint portion 25-1,25-2 ..., 2 sub-bitstream data in 25-(n-1).As shown in Figure 2, the sub-bitstream data SBS to joint portion 25-1 input from the 23-1 of coding portion output ₁with the sub-bitstream data SBS from the 23-2 of coding portion output ₂.Then, joint portion 25-1 is at sub-bitstream data SBS ₁zygote bitstream data SBS afterwards ₂.Then, 25-1 output in joint portion is combined into the sub-bitstream data of.

And, the sub-bitstream data of exporting from joint portion 25-(m-1) to joint portion 25-m (wherein 2≤m≤n-1) input and the sub-bitstream data SBS from the 23-of coding portion (m+1) output _m+1.Then, joint portion 25-m zygote bitstream data SBS after the sub-bitstream data from joint portion 25-(m-1) output _m+1.Then, 25-m output in joint portion is combined into the sub-bitstream data of.Then, joint portion 25-(n-1) output combines all by the bitstream data after the sub-bitstream data of each coding portion generation.This bitstream data is the data after dynamic image data that handling part 12 is obtained is encoded.

In addition, joint portion 25-1,25-2 ..., 25-(n-1) has identical 26S Proteasome Structure and Function, so, below a joint portion 25-m is described.

Joint portion 25-m obtains the residual occupancy volume S notifying from the amount of information control part 35 of the 23-m of coding portion _rwith accept from convergence target leading-out portion 24-m migration time occupancy volume S _trdifference Δ S _r.Then, if this poor Δ S _rbe not 0, joint portion 25-m is at sub-bitstream data and sub-bitstream data SBS from joint portion 25-(m-1) output _m+1between, insert and this poor Δ S _rthe invalid bit column of suitable amount.At antithetical phrase bitstream data SBS _mlast picture while decoding, from standard decoder buffer memory, take out this invalid bit column together with being equivalent to the data of this last picture.Therefore occupancy volume S when, joint portion 25-m can make the occupancy volume of the standard decoder buffer memory that moves the moment and move _trunanimously.In addition, preferably this invalid bit column is the bit column image quality of the picture that will decode not being impacted.For example, this invalid bit column can be the Filler data of Network Abstraction Layer (NAL) form of regulation in H.264MPEG-4AVC.Or this invalid bit column can be the byte of padding specifying in MPEG-2.

Fig. 7 is the figure of the corresponding relation of the sub-bitstream data of migration after being combined by joint portion 25-m that be illustrated in the occupancy volume that occupies of bitstream data in the migration moment in standard decoder buffer memory.The schematic configuration of bitstream data 701 is shown at the upside of Fig. 7.And, at the downside of Fig. 7, curve 710 is shown, the passing of the occupancy volume that these curve 710 these bitstream data 701 of expression occupy in standard decoder buffer memory.About the curve 710 of downside, transverse axis represents the time, and the longitudinal axis represents occupancy volume.And, moment t _trrepresent the migration moment.

As shown in Figure 7, bitstream data 701 has the sub-bitstream data 702 of front side and the sub-bitstream data 703 of rear side.And then bitstream data 701 has invalid bit column 704 between these 2 sub-bitstream data.Wherein, take out valid data 705 corresponding to the last picture that comprises with the sub-bitstream data 702 of front side from standard decoder buffer memory time, as shown in curve 710, the occupancy volume minimizing amount S suitable with these valid data 705 of standard decoder buffer memory _v.At migration moment t _tr, from standard decoder buffer memory, take out invalid bit column 704 together with valid data 705, so the occupancy volume of standard decoder buffer memory further reduces the amount Δ S suitable with invalid bit column 704 _r.Therefore, at migration moment t _trfrom standard decoder buffer memory, take out the occupancy volume after the valid data 705 that the last picture that comprises with the sub-bitstream data 702 of front side is corresponding, occupancy volume S during with the migration of the sub-bitstream data 703 of rear side _trunanimously.

Fig. 8 illustrates the motion flow by the coding processing of the dynamic image data of the computer program control of carrying out on the handling part 12 at moving picture encoding device 1.

As shown in Figure 8, when handling part 12 is obtained after the dynamic image data as coded object, handling part 12 starts coding to be processed.Then, dynamic image data is divided into n subdata (step S101) by the cutting part 22 of handling part 12.Wherein, n is more than 2 integer, equates with the number of coding portion.Then, cutting part 22 is according to the time sequencing of reproducing, and to each coding 23-m of portion, (1≤m≤n) inputs each subdata.

And the convergence target leading-out portion 24-k (1≤k≤n-1) of handling part 12 calculates respectively the sub-bitstream data SBS from being generated by the 23-k of coding portion _kto the sub-bitstream data SBS being generated by the 23-of coding portion (k+1) _k+1occupancy volume and convergence desired value (step S102) when the migration in migration moment of migration.As mentioned above, occupancy volume is at migration sub-bitstream data SBS of moment when this migration _k+1the occupancy volume occupying in standard decoder buffer memory.And then convergence desired value is at migration moment antithetical phrase bitstream data SBS _kwith sub-bitstream data SBS _k+1carry out in conjunction with after the desired value of the occupancy volume that occupies in standard decoder buffer memory of bitstream data BS.Occupancy volume while restraining target leading-out portion 24-k to joint portion 25-k informing removal.And convergence target leading-out portion 24-k is to the 23-k of the coding portion notice convergence desired value of handling part 12.

(1≤m≤n) is according to higher limit and the lower limit of notified convergence desired value or the standard decoder buffer memory set by control part 21, and each picture that inputted subdata is comprised is encoded for each coding 23-m of portion.Then, each coding 23-m of portion generates the sub-bitstream data (step S103) after each picture that subdata is comprised is encoded.Now, each coding 23-k of portion (1≤k≤n-1) generates respectively sub-bitstream data SBS _k, make to combine generated sub-bitstream data SBS _kand SBS _k+1after the residual occupancy volume S that occupies in when migration is engraved in standard decoder buffer memory of bitstream data _rapproach convergence desired value.And the 23-n of coding portion generates sub-bitstream data SBS _n, make sub-bitstream data SBS _nthe occupancy volume occupying in standard decoder buffer memory converges between the maximum permissible value and minimum permissible value of standard decoder buffer memory.Then, the 23-1 of coding portion and the 23-2 of coding portion export the sub-bitstream data generating to joint portion 25-1.And, the 23-m of coding portion (3≤m≤n) export to joint portion 25-(m-1) the sub-bitstream data generating.

Each joint portion 25-k (1≤k≤n-1) obtains the residual occupancy volume S of reality notifying from the amount of information control part 35 of the 23-k of coding portion _rwith accept from convergence target leading-out portion 24-k migration time occupancy volume S _trdifference Δ S _r(step S104).Then, joint portion 25-k, between the sub-bitstream data from joint portion 25-(k-1) output and the sub-bitstream data of accepting from coding portion 23-(k+1), inserts with this and differs from Δ S _rthe invalid bit column of suitable amount.Then, each joint portion 25-k carries out in conjunction with (step S105) these 2 sub-bitstream data.

Then, all the combination of sub-bitstream data is finished in joint portion, thus, generates the bitstream data after dynamic image data is encoded.Then, handling part 12 finishes moving picture encoding processing.

Fig. 9 illustrates the motion flow of the coding processing of the subdata of being undertaken by each coding 23-k of portion (1≤k≤n-1) of the computer program control of carrying out on the handling part 12 at moving picture encoding device 1.

As shown in Figure 9, the buffer memory upper limit/lower limit control part 34 of the 23-k of coding portion is according to the convergence desired value of accepting from convergence target leading-out portion 24-k, and decision is for higher limit and the lower limit (step S201) of the standard decoder buffer memory of each picture of inputted subdata.Now, the lower limit that the buffer memory upper limit/lower limit control part 34 determines as follows for each picture, the position of picture is more approaching by the end of the sub-bitstream data obtaining that inputted subdata is encoded, and lower limit approaches convergence desired value.And lower than convergence desired value in the situation that, the higher limit that the buffer memory upper limit/lower limit control part 34 is revised for each picture, so that this higher limit is higher than convergence desired value in higher limit.Then, the buffer memory upper limit/lower limit control part 34 is notified this higher limit and lower limit to the amount of information control part 35 of the 23-k of coding portion.

Amount of information control part 35 determines the target information amount (step S202) corresponding with coded-bit sendout for each picture.Now, amount of information control part 35 is made as condition, to the sub-bitstream data SBS being generated by the 23-k of coding portion _kwith the sub-bitstream data SBS being generated by the 23-of coding portion (k+1) _k+1carry out in conjunction with after the occupancy volume that occupies in standard decoder buffer memory of bitstream data be included between this higher limit and lower limit.And, amount of information control part 35 is about the picture of having set the lower limit different from lower limit for the previous picture being encoded, calculate the control information amount corresponding with the difference of this lower limit, from target information amount, deduct control information amount, thus, target information amount is proofreaied and correct to (step S203).Then, amount of information control part 35, according to this target information amount, is obtained the quantization parameter (step S204) for determining quantization amplitude.Amount of information control part 35 is notified obtained quantization parameter to orthogonal conversion/quantization unit 32 of the 23-k of coding portion.

Orthogonal conversion/quantization unit 32 carries out orthogonal conversion process for each picture of inputted subdata, obtains frequency signal.Then, the quantization amplitude of orthogonal conversion/quantization unit 32 to determine according to the quantization parameter of notifying from amount of information control part 35, quantizes (step S205) to this frequency signal.Orthogonal conversion/quantization unit 32 is handed to the quantized signal obtaining by this quantification treatment the variable length code portion 33 of the 23-k of coding portion.Then, variable length code portion 33 carries out variable length code processing (step S206) for the quantized signal of accepting from orthogonal conversion/quantization unit 32.The coded-bit row that variable length code portion 33 obtains in conjunction with the each picture comprising for the subdata that is input to coding portion, thus, generate the sub-bitstream data (step S207) after this subdata is encoded.Then, variable length code portion 33 exports the sub-bitstream data (step S208) generating to corresponding joint portion.And variable length code portion 33 is to the actual number of coded bits (step S209) that each picture of generated sub-bitstream data is distributed of amount of information control part 35 notices.Then, the 23-k of coding portion finishes the coding processing of subdata.

In addition, in the 23-n of coding portion, in above-mentioned steps S201, the higher limit of standard decoder buffer memory and lower limit are set as respectively the higher limit and the lower limit itself that are determined by control part 21 by the buffer memory upper limit/lower limit control part 34.And, the processing of omitting above-mentioned steps S203.About other actions, coding portion is same moves with other for the 23-n of coding portion.

In addition, the bitstream data being generated by moving picture encoding device 1 can be for example taking MPEG-2, MPEG-4 or H.264MPEG-4AVC as the data of benchmark.Therefore, can decode to the bitstream data being generated by moving picture encoding device 1 by existing dynamic image decoding device.

As described above, this moving picture encoding device control is for the coded-bit sendout of the sub-bitstream data of the front side in the bitstream data being generated in conjunction with 2 sub-bitstream data by joint portion.Thus, the occupancy volume that the bitstream data after the last picture of the sub-bitstream data of this moving picture encoding device adjustment to front side is decoded migration moment, combination occupies in standard decoder buffer memory.Particularly, to make to move the occupancy volume that occupancy volume that the bitstream data after the combination in moment occupies in standard decoder buffer memory occupies than the sub-bitstream data that was accumulated in the rear side in standard decoder buffer memory before the migration moment high for moving picture encoding device.And then, this moving picture encoding device between the sub-bitstream data of front side and the sub-bitstream data of rear side, the invalid bit column that the last picture that is inserted in the sub-bitstream data to front side takes out while decoding in the lump.Thus, this moving picture encoding device can be eliminated the poor of occupancy volume that the sub-bitstream data of the occupancy volume that occupies in standard decoder buffer memory of bitstream data after the combination in migration moment and rear side occupies in standard decoder buffer memory.Therefore, this moving picture encoding device does not need the simulation of the migration of carrying out occupancy volume, need to after the picture after coding being decoded, again not encode temporarily yet, can meet the regulation of standard decoder buffer memory.Therefore, this moving picture encoding device can be encoded to dynamic image data at short notice.And then, this moving picture encoding device revise gradually in conjunction with after the lower limit of the occupancy volume that occupies in standard decoder buffer memory of bitstream data, so the number of coded bits that each picture that this bitstream data is comprised distributes can sharply not reduce.Therefore, this moving picture encoding device can prevent that the image quality that occurring in dynamic image data is all encodes causes is deteriorated.

In addition, the invention is not restricted to above-mentioned execution mode.For example, this moving picture encoding device also can be encoded to dynamic image data with variable bit rate.In this situation, the occupancy volume that bitstream data occupies in standard decoder buffer memory is not less than the minimum permissible value of this occupancy volume.Therefore,, even if the occupancy volume that the residual occupancy volume of the bitstream data in migration moment occupies than the sub-bitstream data that was accumulated in the rear side in standard decoder buffer memory before the migration moment is many, each colligator also can not insert invalid bit column.

And in the case of moving picture encoding device moving picture encoding standard is according to the rules moved, according to the kind of this standard, the invalid bit column of being inserted by joint portion only has the length of the integral multiple of byte sometimes.On the other hand, the migration moment to 2 sub-bitstream data carry out in conjunction with after bitstream data residual occupancy volume and this 2 sub-bitstream data of in standard decoder buffer memory, occupying in the sub-bitstream data of rear side occupy migration time occupancy volume difference be bit base.Therefore,, the integral multiple that is not byte in this difference, even if invalid bit column is inserted in joint portion, when also cannot making to move the residual occupancy volume in moment and moving, occupancy volume is in full accord.Therefore, the error of occupancy volume during due to the residual occupancy volume after this invalid bit column insertion and migration, the occupancy volume that bitstream data occupies in standard decoder buffer memory may not converge between the maximum permissible value and minimum permissible value of standard decoder buffer memory.

Therefore, the buffer memory upper limit/lower limit control part of preferred each coding portion is set as the lower limit of the standard decoder buffer memory in migration moment than the surplus of the high regulation of convergence desired value.And the buffer memory upper limit/lower limit control part of preferred each coding portion is set as the higher limit of standard decoder buffer memory than the surplus of the low regulation of maximum permissible value of standard decoder buffer memory.The surplus of this regulation is the amount suitable with above-mentioned error, for example, can be 8 bits.And then, the total of above-mentioned error is obtained in preferred each joint portion from previous colligator, determine the length of invalid bit column, make in the total of obtained error, to add the error producing in this joint portion and the absolute value of the value obtaining all the time below the surplus of regulation.

This picture coding device is for various uses.For example, this picture coding device is assembled in dynamic image data issue server, image transfer apparatus, video-telephone system, computer or portable telephone.For example, in the situation that this picture coding device is assembled in dynamic image data issue use server, by its handling part, with together with other signals of obtaining with dynamic image data of voice signal etc. simultaneously, the dynamic image data after the coding being generated by this moving picture encoding device is converted to the data flow based on regulation communication standard.Then the dynamic image data issue server of, having assembled picture coding device is issued this data flow via communication line to dynamic image decoding device.

And, also can be to be recorded in the form in the medium of embodied on computer readable, the computer program of each function of control part, cutting part, coding portion, convergence target leading-out portion and joint portion that the handling part that makes this moving picture encoding device of computer realization has is provided.

Here whole examples and the particular terms enumerated should be interpreted as, taking teaching as object, for helping reader understanding the present invention and the concept of the inventor for advancing this technology to propose, be not limited to structure and this specific example and the condition of the arbitrary examples of this specification relevant with representing Pros and Cons of the present invention.Describe embodiments of the present invention in detail, still, wish to understand, can carry out various changes, displacement and correction to it under the premise without departing from the spirit and scope of the present invention.

Claims (7)

1. a moving picture encoding device, it has:

Storage part, its storage at least comprises the dynamic image data of the 1st subdata and the 2nd subdata, and the 1st subdata and the 2nd subdata comprise respectively multiple pictures;

Handling part, it is by described dynamic image data is encoded, the bitstream data that generated data amount is fewer than the data volume of this dynamic image data; And

Efferent, it exports described bitstream data,

Described handling part is achieved as follows function:

The 1st encoding function, it is so that the 1st occupancy volume becomes the mode of desired value, determine the sendout of the coded-bit of each picture comprising for described the 1st subdata, according to this sendout, described the 1st subdata is encoded, thereby generate the 1st sub-bitstream data;

The 2nd encoding function, it generates the 2nd sub-bitstream data by described the 2nd subdata is encoded;

Binding function, its by after described the 1st sub-bitstream data in conjunction with described the 2nd sub-bitstream data, generate described bitstream data; And

Convergence target export function, with regulation transfer rate transmit described bitstream data to virtual decoder and be accumulated in the buffer memory of this decoder, and when the data suitable with each picture of taking out described bitstream data with official hour interval successively from described buffer memory and being comprised, it determines the described desired value of described the 1st occupancy volume, described the 1st occupancy volume that makes to occupy in described buffer memory in bitstream data described in the 1st time point be at the 1st time point, described the 2nd subdata is encoded to described the 2nd sub-bitstream data that obtains occupies in described buffer memory the 2nd more than occupancy volume, described the 1st time point is encoded moment of the data suitable with last picture that described the 1st sub-bitstream data that obtains comprises to described the 1st subdata for taking out from described buffer memory.

2. moving picture encoding device according to claim 1, wherein,

In described binding function, between described the 1st sub-bitstream data and described the 2nd sub-bitstream data, insert the bit column with the length that the difference that obtains with deduct described the 2nd occupancy volume from described the 1st occupancy volume of described the 1st time point is suitable.

3. moving picture encoding device according to claim 2, wherein,

Described bit column is the invalid bit column that the last picture that in described virtual decoder, described the 1st sub-bitstream data comprised takes out and the decoding of this last picture do not impacted while decoding together with being equivalent to the data of this last picture from described buffer memory.

4. according to the moving picture encoding device described in any one in claim 1～3, wherein,

Described the 1st encoding function has:

Occupancy volume limits value determines function, they are in the time of the data suitable with n picture of taking out from described buffer memory that described the 1st sub-bitstream data comprises, increase the lower limit of the occupancy volume that described bitstream data occupies in described buffer memory, make the position of described n picture approach the last of described the 1st sub-bitstream data, this lower limit approaches described desired value, any one picture that described n picture comprises for described the 1st sub-bitstream data, this n is the value of the 1 last picture number comprising to described the 1st sub-bitstream data; And

Amount of information control function, it in the time taking out described n picture from described buffer memory, so that the occupancy volume that described bitstream data occupies in described buffer memory is mode more than described lower limit, determines the sendout for the coded-bit of described n picture.

5. moving picture encoding device according to claim 4, wherein,

Determine in function at described occupancy volume limits value, in described virtual cache, accumulated that described the 1st sub-bitstream data comprises data suitable with m picture time, the pre-determined higher limit of the occupancy volume occupying in described virtual cache in described bitstream data is less than described desired value, this higher limit is modified to higher than described desired value, any one picture that described m picture comprises for described the 1st sub-bitstream data, this m is the value of the 1 last picture number comprising to described the 1st sub-bitstream data.

6. moving picture encoding device according to claim 4, wherein,

Described the 1st encoding function also has:

Quantization function, multiple pictures that it comprises for described the 1st sub-bitstream data respectively, the signal corresponding with each pixel of this picture carried out to orthogonal conversion, thereby generated frequency signal, the quantization amplitude that sendout is larger, this frequency signal is less with described coded-bit, quantizes described frequency signal; And

Variable length code function, it generates described the 1st sub-bitstream data by the described frequency signal after described quantification is carried out to variable length code,

In described amount of information control function, from the base quantity of the coded-bit to described n picture distribution in the time that the 1st lower limit equates with the 2nd lower limit, deduct the correcting value of the difference that is equivalent to described the 1st lower limit and described the 2nd lower limit, thereby determine the sendout for the described coded-bit of described n picture, wherein, described the 1st lower limit is the described lower limit of the described n picture that comprises for described the 1st sub-bitstream data, and described the 2nd lower limit is the lower limit for the previous picture being encoded of this n picture.

7. a motion image encoding method, includes respectively the 1st subdata of multiple pictures and the dynamic image data of the 2nd subdata is encoded at least comprising, thus the generated data amount bitstream data fewer than the data volume of this dynamic image data, wherein,

So that the 1st occupancy volume becomes the mode of described desired value, determine the sendout of the coded-bit of each picture comprising for described the 1st subdata, according to this sendout, described the 1st subdata is encoded, thereby generate the 1st sub-bitstream data,

By described the 2nd subdata is encoded, generate the 2nd sub-bitstream data,

By after described the 1st sub-bitstream data in conjunction with described the 2nd sub-bitstream data, generate described bitstream data,

With regulation transfer rate transmit described bitstream data to virtual decoder and be accumulated in the buffer memory of this decoder, and when the data suitable with each picture of taking out described bitstream data with official hour interval successively from described buffer memory and being comprised, determine the desired value of described the 1st occupancy volume, described the 1st occupancy volume that makes to occupy in described buffer memory in bitstream data described in the 1st time point be at the 1st time point, described the 2nd subdata is encoded to the 2nd sub-bitstream data that obtains occupies in described buffer memory the 2nd more than occupancy volume, described the 1st time point is encoded moment of the data suitable with last picture that the 1st sub-bitstream data that obtains comprises to described the 1st subdata for taking out from described buffer memory.